CN108615031A - Heart sound filtering method based on threshold value wavelet transformation - Google Patents
Heart sound filtering method based on threshold value wavelet transformation Download PDFInfo
- Publication number
- CN108615031A CN108615031A CN201810500532.XA CN201810500532A CN108615031A CN 108615031 A CN108615031 A CN 108615031A CN 201810500532 A CN201810500532 A CN 201810500532A CN 108615031 A CN108615031 A CN 108615031A
- Authority
- CN
- China
- Prior art keywords
- signal
- cardiechema signals
- threshold value
- filtering
- heart sound
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
- G06F2218/06—Denoising by applying a scale-space analysis, e.g. using wavelet analysis
Abstract
The invention belongs to digital information processings and medical domain, and in particular to a kind of cardiechema signals filtering method based on threshold value wavelet transformation.Untreated cardiechema signals audio is obtained from MIT BIT arrhythmia cordis databases and seeks the data such as Signal to Noise Ratio (SNR) and root-mean-square error RMSE to it, after the cardiechema signals of gained are carried out wavelet decomposition, seeks each layer details and corresponding detail coefficients.Threshold process is carried out to each layer coefficients again.Signal after reconstruction filtering, and it generates audio after filtering heart sound audio file seeks filtering and seeks the data such as Signal to Noise Ratio (SNR) and root-mean-square error RMSE, compare the front and back data variation of filtering, because the signal-to-noise ratio of the front and back signal of filtering has very big promotion, and root-mean-square error is obviously reduced, therefore have positive effect to cardiechema signals progress wavelet filter.
Description
Technical field
The invention belongs to digital information processings and medical domain, and in particular to a kind of heart sound letter based on threshold value wavelet transformation
Number filtering method.
Background technology
Detection for cardiechema signals, wave band residing for useful signal, that is, cardiechema signals are 3 ~ 40Hz；And heart sound includes process
In, measured's local environment noise, itself respiratory noise, myoelectricity interference frequency and with its own institute band the general frequency of noise exist
Between 50 ~ 2000Hz.One normal cardiac cycle includes the composition of four heart sound, is often divided with S1, S2, S3, S4,
Middle S1 is in ventricular contraction period, and S2 is in ventricular diastole period, and S3 is happened at after S2 0.1~0.2 second, its main feature is that frequency compared with
S1, S2 are low, this is because blood quick washing core indoor wall in this stage, causes ventricle and valve vibration of membrane, due to S3 energy
It measures relatively low, it is more difficult to be transmitted to body surface, therefore can be only monitored to when tested personnel is children.S4 is drawn by atrial contraction
It rises, also referred to as atrial heart sound.S3, S4 intensity are low under normal conditions, the general signal only considered between S1, S2.
For wavelet transformation（Wavelet transfer, WT）For, it can be by the partial transformation to time domain, to signal
Local message effectively extracted.Meanwhile wavelet analysis has preferable adaptivity to time domain and frequency-region signal, it can be right
Signal carries out more convenient and fine processing.
It can be seen that for untreated cardiechema signals, cardiechema signals concentrate on low-frequency range, and noise is largely distributed in medium-high frequency
Section.The high band of signal is ideally separated off while retaining low-frequency range as far as possible by therefore denoising.Wavelet transformation exists
The height frequency range of signal is detached and is reached with this by the characteristics of to that can play its high-resolution in the time frequency analysis of signal well
Remove the purpose of noise.
The basic procedure of Wavelet Denoising Method introduced below：Suitable wavelet basis function is determined first and input signal is decomposed
The number of plies, wavelet decomposition then is carried out to pending signals and associated noises.The threshold value decomposed to every layer is sought and is recorded, then ties
Each layer details gone out to signal decomposition is closed, denoising is carried out to every layer signal that decomposition removes respectively.Finally according to Decomposition order
And relevant details reconstructs to obtain filtered signal, is filtered to original signal to reach.
Invention content
Present invention proposition is a kind of to carry out quick filter processing to reach using threshold value wavelet transformation to heart sound original signal
It makes an uproar and improves the method that heart sound judges to sort out accuracy.
The present invention has following steps：
1）Untreated cardiechema signals audio is obtained from MIT-BIT arrhythmia cordis databases and signal-to-noise ratio (SNR) is sought to it
With the data such as root-mean-square error (RMSE)；
2）Cardiechema signals obtained by step 1 are subjected to wavelet decomposition；
3）Corresponding detail coefficients are sought to each layer details of wavelet decomposition in step 2；
4）Threshold process is carried out to each layer coefficients of step 3；
5）Signal after reconstruction filtering, and generate filtering heart sound audio file；
6）The data such as SNR and RMSE are sought to audio after filtering obtained by step 5, the front and back data variation of filtering is compared and examines filtering effect
Fruit.
Description of the drawings
1）Fig. 1 flow charts；
2）Fig. 2 original signals part time frequency signal；
3）Fig. 3 original signals part frequency-region signal；
4）Fig. 4 reconstruct six layers of approximation coefficient；
5）Mono- layer of decomposition details of Fig. 5；
6）Bis- layers of decomposition details of Fig. 6；
7）Tri- layers of decomposition details of Fig. 7；
8）Tetra- layers of decomposition details of Fig. 8；
9）Five layers of decomposition details of Fig. 9；
10）Six layers of decomposition details of Figure 10；
11）Figure 11 local time frequency signals after being filtered；
12）Figure 12 local frequency-region signals after being filtered；
13）The front and back local signal time domain comparison diagram of Figure 13 filtering.
Specific implementation mode
Below in conjunction with attached drawing 1, the invention will be further described.
1）Obtain untreated cardiechema signals audio from MIT-BIT arrhythmia cordis databases and it is sought SNR and
The data such as RMSE；
Specially：Untreated cardiechema signals audio is obtained from MIT-BIT arrhythmia cordis databases, if the mould of original signal
Type is
（1）
Wherein,For signals and associated noises,For actual signal,For noise.
It is incumbent to take a cardiechema signals audio file, seek its SNR and RMES using MATLAB programmings.
2）Cardiechema signals obtained by step 1 are subjected to wavelet decomposition；
Specially：For cardiechema signals, useful part concentrates on low frequency part, and high frequency section is myoelectricity interference, detection machinery
Deng the noise brought into.Therefore, after heart sound being carried out wavelet decomposition, can cardiechema signals be carried out with the decomposition of certain number of plies, then
It is upper at every layer to be handled using corresponding threshold value, achieve the purpose that denoising.For wavelet basis function sym6, scaling function with
The characteristic wave bands of cardiechema signals are very close to and its mathematical characteristic is orthogonal but not Striking symmetry, is well suited for for cardiechema signals
Carry out wavelet decomposition.Therefore this patent selects sym6 wavelet basis functions to carry out 6 layers of wavelet decomposition to cardiechema signals.
3）Corresponding detail coefficients are sought to each layer details of wavelet decomposition in step 2；
Specially：To seeking six layers of approximation of signal in 6 layers of decomposition of gained in step 2And its 1 ~ 6 layer of details,
Middle j indicates that the number of plies decomposed, i indicate i-th of data.
4）Threshold process is carried out to each layer coefficients of step 3；
Specially：Use calculated threshold valueTojThe detail section of layerIt is handled.It is as follows to handle formula：
(2)
EvenWhen, then retain the details；IfWhen, then propose the details.It is right with thisIt is sieved
Choosing is handled.
In general, threshold value is selected according to the signal-to-noise ratio of original signal.Each layer threshold value in this patentTo call
Wdcbm functions in MATLAB acquire.
5）Signal after reconstruction filtering, and generate filtering heart sound audio file；
Specially：By 6 layers of approximation and 1 ~ 6 layer of detailsIt is reconstructed using algorithm, obtains filtered cardiechema signals.
6）The data such as SNR and RMSE are sought to audio after filtering obtained by step 5, the front and back data variation of filtering is compared and examines filter
Wave effect；
Specially：Seeking for SNR and RMSE is carried out to cardiechema signals after the filtering obtained by step 5, with the original signal obtained by step 1
SNR and RMSE are compared, and examine whether filter effect reaches.
It will be further detailed below by example.By the optional cardiechema signals of MIT-BIT arrhythmia cordis databases
A0007.wav, local temporal performance plot and local frequency domain characteristic figure are as shown in Figure 2,3.It is used in combination MATLAB programmings to seek its SNR
And RMSE.Sym6 wavelet basis functions are now selected to carry out 6 layers of wavelet decomposition to it, one layer to six layers of Hierarchical Detailed is respectively such as Fig. 5
Shown in Figure 10.MATLAB programmings are recycled to seek six layers of approximation of signal respectivelyAnd its 1 ~ 6 layer of details.It calls
Wdcbm functions in MATLAB acquire each layer threshold value, use calculated threshold valueTojThe detail section of layerInto
Row processing.By 6 layers of approximation and 1 ~ 6 layer of detailsIt is reconstructed using algorithm, obtains filtered cardiechema signals, filtering signal
Time domain, frequency domain characteristic figure be respectively Figure 11, Figure 12.The energy for comparing signal institute band before and after visible filtering by Fig. 3 and Figure 12 has
Weakened, is because the noise in signal has largely been filtered out.SNR and RMSE is carried out to cardiechema signals after filtering to ask
It takes, is compared such as table 1 with the obtained original signal SNR and RMSE of original signal a0007.wav.
| Original signal | Filtering signal | |
| SNR | 6.1578 | 15.2433 |
| RMSE | 0.4459 |
Table 1
By table 1 and the front and back comparison diagram of filtering, the signal-to-noise ratio for filtering front and back signal has very big promotion, and root-mean-square error obviously subtracts
It is small.It is clearly visible again by Figure 13, filtered signal noise is filtered out by apparent, clearly can distinguish S1, S2 two
A stage, and signal is more smooth, offers convenience to subsequent signal processing.Therefore wavelet filter is carried out to cardiechema signals
There is positive effect.
Claims (7)
1. carrying out quick filter processing to heart sound original signal to reach denoising and retain cardiechema signals based on threshold value wavelet transformation
The method of effective information, which is characterized in that include the following steps：
Obtain untreated cardiechema signals audio from MIT-BIT arrhythmia cordis databases and it is sought Signal to Noise Ratio (SNR) and
The data such as square error RMSE；
Cardiechema signals obtained by step 1 are subjected to wavelet decomposition；
Corresponding detail coefficients are sought to each layer details of wavelet decomposition in step 2；
Threshold process is carried out to each layer coefficients of step 3；
Signal after reconstruction filtering, and generate filtering heart sound audio file；
The data such as Signal to Noise Ratio (SNR) and root-mean-square error RMSE are sought to audio after filtering obtained by step 5, compare the front and back data of filtering
Filter effect is examined in variation；
It is according to claim 1 that quick filter processing is carried out to reach to heart sound original signal based on threshold value wavelet transformation
The method for making an uproar and retaining cardiechema signals effective information, which is characterized in that the step 1 is specially：
Untreated cardiechema signals audio is obtained from MIT-BIT arrhythmia cordis databases, if the model of original signal is
（1）
Wherein,For signals and associated noises,For actual signal,For noise；
It is incumbent to take a cardiechema signals audio file, seek its Signal to Noise Ratio (SNR) and root-mean-square error using MATLAB programmings
RMES。
2. according to claim 1 carry out quick filter processing to reach based on threshold value wavelet transformation to heart sound original signal
Denoising and the method for retaining cardiechema signals effective information, which is characterized in that the step 2 is specially：Heart sound is subjected to small wavelength-division
Xie Hou, for cardiechema signals, useful part concentrates on low frequency part, and high frequency section is institutes' bands such as myoelectricity interference, detection machinery
Therefore the noise entered can carry out cardiechema signals the decomposition of certain number of plies, then upper using at corresponding threshold value at every layer
Reason, achievees the purpose that denoising, Decomposition order is more, and high frequency section is removed more, but the number of plies excessively can be useful by part
Signal is rejected together, that is, is decomposed excessively, for wavelet basis function sym6, the characteristic wave bands of scaling function and cardiechema signals connect very much
Closely, and its mathematical characteristic is orthogonal but not Striking symmetry, is well suited for for carrying out wavelet decomposition to cardiechema signals；Therefore this patent
Sym6 wavelet basis functions are selected to carry out 6 layers of wavelet decomposition to cardiechema signals.
3. according to claim 1 carry out quick filter processing to reach based on threshold value wavelet transformation to heart sound original signal
Denoising and the method for retaining cardiechema signals effective information, which is characterized in that the step 3 is specially：To 6 of gained in step 2
Layer seeks six layers of approximation of signal in decomposingAnd its 1 ~ 6 layer of details, the number of plies that wherein j expressions are decomposed, i-th of i expressions
Data.
4. according to claim 1 carry out quick filter processing to reach based on threshold value wavelet transformation to heart sound original signal
Denoising and the method for retaining cardiechema signals effective information, which is characterized in that the step 4 is specially：Use calculated threshold valueTojThe detail section of layerIt is handled, processing formula is as follows：
(2)
I.e.When, it enables；When, it enables.
5. right with thisScreening Treatment is carried out, in general, threshold value is selected by the signal-to-noise ratio of original signal, each in this patent
Layer threshold valueIt is acquired for the wdcbm functions in Calling MATLAB.
6. according to claim 1 carry out quick filter processing to reach based on threshold value wavelet transformation to heart sound original signal
Denoising and the method for retaining cardiechema signals effective information, which is characterized in that the step 5 is specially：By 6 layers of approximation and 1 ~ 6 layer
DetailsIt is reconstructed using algorithm, obtains filtered cardiechema signals.
7. according to claim 1 carry out quick filter processing to reach based on threshold value wavelet transformation to heart sound original signal
Denoising and the method for retaining cardiechema signals effective information, which is characterized in that the step 6 is specially：To the filtering obtained by step 5
Cardiechema signals carry out seeking for Signal to Noise Ratio (SNR) and root-mean-square error RMSE afterwards, with obtained by step 1 original signal Signal to Noise Ratio (SNR) and
Root-mean-square error RMSE is compared, and examines whether filter effect reaches.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810500532.XA CN108615031A (en) | 2018-05-23 | 2018-05-23 | Heart sound filtering method based on threshold value wavelet transformation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810500532.XA CN108615031A (en) | 2018-05-23 | 2018-05-23 | Heart sound filtering method based on threshold value wavelet transformation |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108615031A true CN108615031A (en) | 2018-10-02 |
Family
ID=63663597
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810500532.XA Pending CN108615031A (en) | 2018-05-23 | 2018-05-23 | Heart sound filtering method based on threshold value wavelet transformation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108615031A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102270270A (en) * | 2011-04-28 | 2011-12-07 | 东北大学 | Remote medical auscultation and consultation system |
| CN103961092A (en) * | 2014-05-09 | 2014-08-06 | 杭州电子科技大学 | Electroencephalogram signal denoising method based on self-adaption threshold processing |
| US20140270519A1 (en) * | 2008-10-17 | 2014-09-18 | Samsung Electronics Co., Ltd. | Image processing apparatus and method of providing high sensitive color images |
| CN105913393A (en) * | 2016-04-08 | 2016-08-31 | 暨南大学 | Self-adaptive wavelet threshold image de-noising algorithm and device |
-
2018
- 2018-05-23 CN CN201810500532.XA patent/CN108615031A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140270519A1 (en) * | 2008-10-17 | 2014-09-18 | Samsung Electronics Co., Ltd. | Image processing apparatus and method of providing high sensitive color images |
| CN102270270A (en) * | 2011-04-28 | 2011-12-07 | 东北大学 | Remote medical auscultation and consultation system |
| CN103961092A (en) * | 2014-05-09 | 2014-08-06 | 杭州电子科技大学 | Electroencephalogram signal denoising method based on self-adaption threshold processing |
| CN105913393A (en) * | 2016-04-08 | 2016-08-31 | 暨南大学 | Self-adaptive wavelet threshold image de-noising algorithm and device |
Non-Patent Citations (2)
| Title |
|---|
| 郭兴明 等: "小波包与混沌集成的心音特征提取及分类识别", 《仪器仪表学报》 * |
| 陈远贵 等: "基于一种新的小波阈值函数的心音信号去噪", 《计算机仿真》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108158573B (en) | Electrocardiosignal noise reduction method based on adaptive threshold wavelet transformation | |
| Alfaouri et al. | ECG signal denoising by wavelet transform thresholding | |
| Kania et al. | Wavelet denoising for multi-lead high resolution ECG signals | |
| CN103405227B (en) | Double-layer morphological filter based electrocardiosignal preprocessing method | |
| CN103961092B (en) | EEG Noise Cancellation based on adaptive thresholding | |
| Kumar et al. | Performance analysis of different filters for power line interface reduction in ECG signal | |
| Biswas et al. | Removing power line interference from ECG signal using adaptive filter and notch filter | |
| CN110151175A (en) | Surface electromyogram signal noise-eliminating method based on CEEMD and improvement wavelet threshold | |
| Hadjileontiadis | Wavelet-based enhancement of lung and bowel sounds using fractal dimension thresholding-Part II: Application results | |
| Tang et al. | ECG de-noising based on empirical mode decomposition | |
| He et al. | Noise reduction of ECG signals through genetic optimized wavelet threshold filtering | |
| Sharma et al. | Multiscale wavelet energies and relative energy based denoising of ECG signal | |
| Jain et al. | An adaptive method for shrinking of wavelet coefficients for phonocardiogram denoising | |
| US9167971B2 (en) | Heart murmur extraction and heart impairments identification using fuzzy controller | |
| Manikandan et al. | Automatic SWT based QRS detection using weighted subbands and Shannon energy peak amplification for ECG signal analysis devices | |
| Priyadharsini et al. | An Efficient method for the removal of ECG artifact from measured EEG Signal using PSO algorithm | |
| CN108615031A (en) | Heart sound filtering method based on threshold value wavelet transformation | |
| CN110101383B (en) | Wavelet energy-based electrocardiosignal denoising algorithm | |
| González–rodríguez et al. | Robust denoising of phonocardiogram signals using time-frequency analysis and U-Nets | |
| Agrawal et al. | Wavelet subband dependent thresholding for denoising of phonocardiographic signals | |
| Figoń et al. | ECG signal quality improvement techniques | |
| Yu et al. | A new method without reference channels used for ventricular fibrillation detection during cardiopulmonary resuscitation | |
| Rahul et al. | Baseline correction of ECG using regression estimation method | |
| Saxena et al. | Optimisation of thresholding techniques in de-noising of ECG signals | |
| Lee et al. | A hybrid approach of wavelet-based total variation and Wiener filter to denoise adventitious lung sound signal for an accurate assessment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20181002 |